Reproducing complex simulations of economic impacts of climate change with lower-cost emulators

In this study, we developed various emulators that mimic the behaviours of simulation models, namely economic models coupled with bio/physical-process-based impact models, by statistical regression techniques

Jun'ya Takakura; Shinichiro Fujimori; Kiyoshi Takahashi; Naota Hanasaki; Tomoko Hasegawa; Yukiko Hirabayashi; Yasushi Honda; Toshichika Iizumi; Chan Park; Makoto Tamura; Yasuaki Hijioka

2021

Scholarcy highlights

  • Climate change has diverse impacts on society and a wide range of sectors, and these impacts should be quantitatively evaluated to manage overall risks
  • We report results for r in the main text since the three metrics, and RMSE to standard deviation) varied almost parallelly, and systematic errors were nearly negligible for all conditions
  • Summarized results beyond r are available in Tables S4 to S13 in the Supplement, and individual values for all sectors and regions are available as electronic supplementary material
  • This implies that using simple damage functions can be useful to grasp the rough picture of economic impacts of climate change
  • The effects of including regional climate conditions are distinct in the economic impacts of thermal power generation and fluvial flooding, whose impacts are strongly affected by local precipitation and river flows
  • Applying ANNbased techniques to economic impact emulation is one of the novelties of this study, and we have demonstrated that these techniques can improve the performance of the emulations
  • If we only focus on the aggregated economic impacts of climate change, a simple damage function which only leverages global mean temperature is worth using provided that we regard the original simulation results as valid
  • If the emulators will be used to estimate economic impacts under different representative concentration pathways or substantially different emission pathways, which are not included in the training data, cross-validation by RCPs can be effective to estimate the actual performance of the emulators in that situation

Need more features? Save interactive summary cards to your Scholarcy Library.